Over the past 25 years, our ability to discover and characterize viral agents has steadily improved, leading to a constant flow of discovery of novel plant viruses as testified by the literature and by the constant increase in the number of viral species described in successive reports of the International Committee for the Taxonomy of Viruses . The development of next generation sequencing (NGS) techniques promises to increase the rate at which novel plant viruses will be discovered in coming years [2, 3]. At the same time, our ability to unambiguously establish the contribution of newly characterized viral agents to particular plant diseases has not improved. The fulfilling of Koch's postulates has been modified by L. Bos to be adapted to viruses, and represents a fundamental point in plant virology . With the application of these postulates, the role of many viruses in diseases has been deciphered. But for many other plant viruses, technical problems in the identification of alternative herbaceous hosts, in purification or in experimental transmission have prevented the analysis of their contribution to particular diseases . This is especially true for viruses affecting vegetatively propagated crops [5–7], which often have the added disadvantage of being frequently mixed infections . Thus, for many viruses, the demonstration of their involvement in a given disease has not been completed, but has only been postulated on the basis of an association with symptomatic plants [see for example [9, 10]].
One strategy to bypass the problems encountered with fulfilling of Koch's postulates involves the use of full-length cDNAs clones (FL-cDNAs) (or DNA clones in the case of DNA viruses) from which infectious RNA transcripts can be obtained in vitro or in vivo . However, the construction of infectious FL-cDNAs is still often complicated and time-consuming for many reasons: the difficulty to optimize a standardized protocol for all viruses, the necessity of a perfect junction of the promoter and 5' end of the viral sequence, the difficulty to clone large cDNA molecules and the frequent instability of such clones .
These difficulties have largely limited the use of FL-cDNAs to studies on reverse genetics of well characterized viruses, which have provided access to valuable information on the expression of viral genomes, their replication and mechanisms involved in the infection cycle. They also provided further insight on the functions of different viral proteins or the mechanisms of interaction between viruses and their host plant(s) or vector(s).
However, despite their potential to address such questions, the use of infectious FL-cDNAs to confirm or refute etiological hypotheses has been rather limited [12–15]. In a recent example, the construction of an agroinfiltrable FL-cDNA clone of Citrus leaf blotch virus (CLBV) allowed the demonstration that CLBV is the causal agent of the Dweet mottle disease and that in single infections it does not cause the bud union crease disease . An example of the widespread use of infectious constructs for etiology studies is in the Geminiviridae family, for which efficient techniques exist for the development of both cloned or uncloned infectious DNA constructs [17, 18]. However, there are additional technical difficulties when working with RNA viruses that are responsible for the limited use of FL-cDNAs in etiology studies of RNA plant viruses.
Simplified strategies for the easier and faster development of infectious FL-cDNA for etiology studies of plant viruses should have a number of desirable properties. First, is the ability to use total nucleic acids (TNA) extracts from infected plants as template for cDNA synthesis [12, 19], rather than purified viral RNA as this would bypass the need to propagate and purify the virus. Second, is the ability to use long distance PCR  to amplify the viral genomes as single, large PCR fragments, a technique that has been used rarely for genomes longer than 7 kb [12, 19, 21–23]. In a number of situations, cloning of the infectious FL-cDNA may not be necessary to validate an etiology hypothesis, so that the ability to infect plants using uncloned PCR products is also of potential interest [24–26]. Lastly, when cloning of FL-cDNAs is used, techniques that facilitate the cloning of long PCR fragments or the one-step assembly of complex constructions would be of great interest. One little used strategy with such a potential is the use of the efficient homologous recombination machinery of the yeast Saccharomyces cerevisiae. Until recently, the application of this system has been limited to yeast genetics and to the construction of plasmids and yeast artificial chromosomes (YACs) . The full power of this approach has been demonstrated recently by the assembly of 25 overlapping DNA fragments to generate a synthetic mycoplasma genome in a single step . In virology, the application of this strategy has been used as an alternative to difficult classical cloning in Escherichia coli, as in the case of Dengue virus type 2, where three cDNA fragments of the virus were assembled by homologous recombination in yeast to generate an infectious FL-cDNA . The fact that recombination is very efficient, even with short, 20-30 nucleotides-long overlap regions between fragments created using PCR primers has facilitated the construction of recombinant viral genomes [30, 31].
In the present study, Apple chlorotic leaf spot virus (ACLSV), the type species of the genus Trichovirus within the family Betaflexiviridae [1, 32, 33], was used as a model system for the development of approaches that fulfill some of the above criteria for improved preparation of infectious FL-cDNAs. The genomic RNA of ACLSV is about 7.55 kb in length [34, 35] and an infectious FL-cDNA for a Japanese isolate from apple (P-205) under the control of the CaMV 35S promoter has been constructed . We report on the long distance PCR amplification from TNA extracts of infectious FL-cDNAs under the control of the T7 promoter. We also show that the yeast homologous recombination system permitted the efficient cloning of such large FL-cDNAs and the simultaneous one-step tailoring of a ternary yeast-E. coli-Agrobacterium tumefaciens shuttle vectors allowing efficient infection of plants by agroinfiltration.